Across power generation, natural gas, electric utilities, and renewables, the same pattern shows up in every AI conversation we have with clients: a short list of things that are clearly working, a longer list of challenges, a few things that are not working and will not, and a real opportunity that most organizations have not yet organized themselves to take. On Tuesday, September 15, I bring that pattern to the technology panel at the 38th Annual LDC Gas Forum Mid-Continent in Chicago, "Technology. Not A Threat. An Opportunity." This post is my working notes ahead of the room.
What the Chicago Panel Is For
The panel title says it plainly: leveraging innovative technology solutions to seize opportunities and overcome challenges. Cleve Hogarth of Cleveland Advisory moderates five of us who build or deploy technology across the natural gas commercial value chain, and the brief asks for practical methods rather than a product tour. Four questions organize the discussion: what is working, what is a challenge, what is not working, and how an organization actually leverages the opportunity.
What follows is drawn from what we see across client engagements in power generation, natural gas, electric utilities, and renewables. No client is named and no result is quoted. The patterns are the point, and they hold across every one of those verticals.
What Is Working With AI in Energy Right Now
The applications that are producing results share a shape: the input is language or pattern, the output is a first pass, and a qualified person checks it before it moves. Where that shape holds, AI is working today.
- Intelligence briefs. Market, regulatory, and competitive monitoring turned into a daily or weekly brief. For a gas desk that is the storage report, weather, basis, and pipeline notices. For a utility it is commission dockets, rate cases, and resource plan filings. For a renewables developer it is interconnection queue movement and tariff or incentive changes. For a generator it is ISO market notices and fuel. The sources differ. The workflow is identical.
- First-pass document work. Contracts, RFP responses, specifications, bid packages, and proposals drafted from an approved library and a playbook of preferred terms, then reviewed by the people who own them. This is the single most common win across every vertical we work in.
- Low-code and no-code agents for repeatable operational tasks. Scheduled AI agents that monitor a source, summarize what changed, route it to the right person, and file it. Document classification, meeting notes turned into action logs, invoice and filing intake. Increasingly these are built by the analyst or coordinator who owns the process, on a low-code platform, in an afternoon. Not glamorous, and exactly where hours come back first.
- Stakeholder and customer communication drafts. Rate case explainers, outage and restoration updates, community notices for a pipeline or solar project, investor and board updates. AI produces the plain-English first draft in minutes. Communications and legal still approve every word.
- Data cleanup and reporting. Turning spreadsheets, PDFs, and scanned records into structured data, then into the dashboards and regulatory reports that used to consume analyst weeks each quarter.
- Being found by the people who are researching you. Engineers, procurement teams, and buyers now research vendors and counterparties through search and AI engines before the first call. Organizations that have organized their public information for that behavior are getting into more conversations. Organizations that have not are being left out of shortlists they never knew existed.
What Is a Challenge
None of the challenges we see are about the models. They are about the organizations deploying them.
- Data readiness. The tool is fine. The data underneath it is inconsistent, siloed, or trapped in PDFs, and every output inherits the mess. This is the number one reason a promising pilot produces an unimpressive result.
- Integration debt. New AI capability that does not connect to the systems of record, so staff end up doing double entry and trust erodes within a quarter.
- Security and confidentiality. Contracts, customer data, and anything adjacent to critical infrastructure belong inside an enterprise deployment behind your firewall, with a written data-handling policy. Most organizations have the concern. Fewer have the policy.
- Accountability in a regulated business. A commission, an ISO, or an auditor does not accept "the model said so." Someone signs, and that person has to understand and be able to defend the reasoning.
- Change management and skills. Teams asked to adopt a new platform every quarter stop adopting anything. The scarce skill is not prompting. It is knowing what a wrong answer looks like in your market.
- Measuring return. Most organizations measure "AI adoption." The organizations that keep funding it measure hours returned and cycle time reduced, per workflow.
- Vendor noise. Every platform in the energy stack now carries an AI label. Separating a real capability from a renamed feature takes time most teams do not have.
What Is Not Working
Some of this is uncomfortable, and it is the part of the panel I expect to be most useful.
- Pilots without owners. A tool proves itself with one team and stops there, because nobody owned the rollout. We see this more than anything else.
- Strategy without a workflow. An AI strategy deck, an AI committee that meets quarterly, and no single process that runs differently on Monday morning.
- Consumer chatbots for confidential work. Staff pasting contract terms, customer records, or operational data into a public tool because nobody gave them a sanctioned one.
- Handing AI the decision. Negotiation positions, dispatch and operational calls, safety-critical judgments, compliance sign-off. AI can prepare every one of these. It should decide none of them, and the organizations that have tested that line have pulled back.
- Buying the platform before fixing the data. A six-figure deployment on top of the same broken spreadsheets, producing faster versions of the same wrong numbers.
- Treating it as an IT project. The wins are in commercial, regulatory, communications, and operations workflows. When IT owns the whole program, those teams never get a workflow of their own.
The Low-Code and No-Code Agent Opportunity
The biggest shift in the last twelve months is not a smarter model. It is that building an agent no longer requires a developer. An agent, in plain terms, is a workflow that watches for a trigger, gathers what it needs, reasons over it, takes a defined action, and stops for a person at the point you tell it to. Low-code and no-code platforms now let the person who owns a process build that workflow with a visual builder and a written instruction instead of a software project.
That changes who gets to leverage the opportunity. The gas supply coordinator who knows exactly which pipeline notices matter can build the agent that watches for them. The regulatory analyst who reads every commission docket can build the one that summarizes new filings each morning. The proposal manager can build the intake agent that classifies incoming RFPs against the go/no-go criteria. None of them need to wait for IT's roadmap.
The platforms doing this work today include Make.com, n8n, and Zapier on the automation side; Microsoft Copilot Studio inside the Microsoft 365 environment most utilities already run; and the agent builders now offered directly by OpenAI, Anthropic, and Google. Each has free tiers or free training, and each connects to the systems commercial teams already use: email, spreadsheets, document stores, calendars, and the web.
Three rules keep this from becoming the next shadow-IT problem:
*Continue reading the full article on this page.*
Key Takeaways
- This insights article shares hands-on strategies for SEO pros, marketing directors, and business owners. Use them to improve organic search and AI visibility across Google, ChatGPT, Perplexity, and other platforms.
- The methods here follow Google E-E-A-T guidelines, Core Web Vitals standards, and GEO best practices for 2026 and beyond.
- Companies that pair technical SEO with strong content, authority link building, and structured data see lasting organic growth. This growth becomes measurable revenue over time.
About the Author: Jason Langella is Founder & Chairman at SEO Agency USA, delivering enterprise SEO and AI visibility strategies for market-leading organizations.